Pairwise benchmark snapshot · Aug 16, 2026
In this head-to-head showdown, GPT-5.6 Luna delivers higher overall intelligence and human-preferred responses (Elo 1,466 vs 1,362), while Claude 3.7 Sonnet leads in SWE-bench software engineering benchmarks (70.3%), while GPT-5.6 Luna is more budget-friendly at $1.2/1M per 1M output tokens. Review the complete breakdown below to determine which model best fits your performance and budget requirements.
9 direct benchmark disciplines evaluated across capability, speed, and cost
Hybrid-reasoning Sonnet from 2025. Kept for historical compare pages.
OpenAI 5.6 fast/cheap tier. Official API $0.20/$1.20 per 1M after the Jul 30 80% Luna cut.
Simulate accuracy gains vs added response time & token cost
Reasoning models “think before answering” by generating internal reasoning tokens. Higher effort improves math, coding, and logic accuracy, but increases response delay and token costs.
Δ 104 Arena Elo pts
188 tok/s
$1.2 / 1M output
Relative percentile scores computed across all active models in the benchmark catalog.
Comparing top Western standard models with China's leading frontier rival across 6 skill dimensions. Tap any spoke or dot to inspect.
Category wins across reasoning intelligence, generation speed, and token cost.
Preference Elo, Coding proficiency, SWE-bench & LiveBench accuracy
Generation throughput and time to first token responsiveness
Cost per million tokens and max context window length
| Benchmark | Claude 3.7 Sonnet | GPT-5.6 Luna | Advantage Delta |
|---|---|---|---|
| Preference Elo | +104 | ||
| Coding Elo | — | — | |
| LiveBench | +5.8% | ||
| SWE-bench | +8.5% | ||
| GPQA Diamond | +3.1% | ||
| Time to first token | +152 ms | ||
| Output speed | 85 tok/s | 188 tok/s | +103 tok/s |
| Input price | +$2.8/1M | ||
| Output price | +$13.8/1M | ||
| Context window | +800k |
Simulate monthly production API costs in USD (US Dollar).
GPT-5.6 Luna is estimated to save $305.00/month ($3,660/year).
Higher SWE-bench (70.3%).
Lower output list price ($1.2/1M).
Lower TTFT (88 ms).
Larger window (1M).
Both accept images. Defaulting to the higher-Elo side (GPT-5.6 Luna).
| Target Workload | Recommended Pick | Evaluation Rationale |
|---|---|---|
| Repo / coding agents | Claude 3.7 Sonnet | Higher SWE-bench (70.3%). |
| High-volume chat | GPT-5.6 Luna | Lower output list price ($1.2/1M). |
| Voice / low-latency UI | GPT-5.6 Luna | Lower TTFT (88 ms). |
| Long-document RAG | GPT-5.6 Luna | Larger window (1M). |
| Screenshots / vision | GPT-5.6 Luna | Both accept images. Defaulting to the higher-Elo side (GPT-5.6 Luna). |
Luna is the OpenAI fast/cheap 5.6 SKU after Jul 30. It is not Sol. Sort by price and TTFT, not Elo.
High-volume agents that cannot pay Sol. Compare to Luna before you assume the new cheap SKU wins.
2025 hybrid-reasoning Sonnet. Deprecated for buying, kept for old compare URLs and “what did 3.7 score?”
No comments posted on this matchup yet. Be the first to share an evaluation note!
Dynamic OpenGraph banner generated at /compare/claude-3-7-sonnet-vs-gpt-5-6-luna/opengraph-image
Verified Head-to-Head Telemetry
